MagnificImageUpscalerCreativeNode - ComfyUI Built-in Node Documentation

This node uses the Magnific AI service to upscale and creatively enhance an image. It allows you to guide the enhancement with a text prompt, choose a specific style to optimize for, and control various aspects of the creative process like detail, resemblance to the original, and stylization strength. The node outputs an upscaled image at your chosen factor (2x, 4x, 8x, or 16x), with a maximum output size of 25.3 megapixels.

Inputs

Parameter	Description	Data Type	Required	Range
`image`	The input image to be upscaled and enhanced.	IMAGE	Yes	-
`prompt`	A text description to guide the creative enhancement of the image. This is optional (default: empty).	STRING	No	-
`scale_factor`	The factor by which to upscale the image’s dimensions.	COMBO	Yes	`"2x"` `"4x"` `"8x"` `"16x"`
`optimized_for`	The style or content type to optimize the enhancement process for.	COMBO	Yes	`"standard"` `"soft_portraits"` `"hard_portraits"` `"art_n_illustration"` `"videogame_assets"` `"nature_n_landscapes"` `"films_n_photography"` `"3d_renders"` `"science_fiction_n_horror"`
`creativity`	Controls the level of creative interpretation applied to the image (default: 0).	INT	No	-10 to 10
`hdr`	The level of definition and detail (default: 0).	INT	No	-10 to 10
`resemblance`	The level of resemblance to the original image (default: 0).	INT	No	-10 to 10
`fractality`	The strength of the prompt and intricacy per square pixel (default: 0).	INT	No	-10 to 10
`engine`	The specific AI engine to use for processing. This is an advanced parameter.	COMBO	Yes	`"automatic"` `"magnific_illusio"` `"magnific_sharpy"` `"magnific_sparkle"`
`auto_downscale`	Automatically downscale input image if output would exceed maximum pixel limit (default: False). This is an advanced parameter.	BOOLEAN	No	-

Constraints:

The input image must be exactly one image.
The input image must have a minimum height and width of 160 pixels.
The input image’s aspect ratio must be between 1:3 and 3:1.
The final output size (input dimensions multiplied by the scale_factor) cannot exceed 25,300,000 pixels. If this limit would be exceeded:
- When auto_downscale is enabled, the node automatically reduces the input image size (by no more than 2x) or uses a lower scale_factor so the output stays within the limit.
- When auto_downscale is disabled, the node raises an error.

Outputs

Output Name	Description	Data Type
`image`	The creatively enhanced and upscaled output image.	IMAGE

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 36c38e87f9f1e568c78cf794aeb0a268c6d25d639006eb2cf18ee040d3071ad4

​Inputs

​Outputs

Inputs

Outputs